Serial batching to minimize the weighted number of tardy jobs

The 1|s-batch(∞),r_j|∑w_jU_j scheduling problem takes as input a batch setup time Δ and a set of n jobs, each having a processing time, a release date, a weight, and a due date; the task is to find a sequence of batches that minimizes the weighted number of tardy jobs. This problem was introduced by Hochbaum and Landy (Oper Res Lett 16(2):79–86, 1994); as a wide generalization of Knapsack, it is NP-hard. In this work, we provide a multivariate complexity analysis of the 1|s-batch(∞),r_j|∑w_jU_j problem with respect to several natural parameters. That is, we establish a classification into fixed-parameter tractable and W[1]-hard problems, for parameter combinations of (i) #p = number of distinct processing times, (ii) #w = number of distinct weights, (iii) #d = number of distinct due dates, (iv) #r = number of distinct release dates. Thereby, we significantly extend the work of Hermelin et al. (Ann Oper Res 298:271–287, 2018) who analyzed the parameterized complexity of the non-batch variant of this problem without release dates. As one of our key results, we prove that 1|s-batch(∞),r_j|∑w_jU_j is W[1]-hard parameterized by the number of distinct processing times and distinct due dates. To the best of our knowledge, these are the first parameterized intractability results for scheduling problems with few distinct processing times. Further, we show that 1|s-batch(∞),r_j|∑w_jU_j is fixed-parameter tractable parameterized by #d+#p+#r, and parameterized by #d+#w if there is just a single release date. Both results hold even if the number of jobs per batch is limited by some integer b.